Re: port-sparc/58352: ld: /usr/lib/libubsan.so: undefined reference to `__sync_val_compare_and_swap_8'

[Rin Okuyama <rokuyama.rk%gmail.com@localhost>]

The following reply was made to PR port-sparc/58352; it has been noted by GNATS.

From: Rin Okuyama <rokuyama.rk%gmail.com@localhost>
To: gnats-bugs%netbsd.org@localhost, port-sparc-maintainer%netbsd.org@localhost,
 gnats-admin%netbsd.org@localhost, netbsd-bugs%netbsd.org@localhost, campbell+netbsd%mumble.net@localhost
Cc: 
Subject: Re: port-sparc/58352: ld: /usr/lib/libubsan.so: undefined reference
 to `__sync_val_compare_and_swap_8'
Date: Thu, 20 Jun 2024 00:12:04 +0900

 On 2024/06/19 23:55, Martin Husemann wrote:
 > The following reply was made to PR port-sparc/58352; it has been noted by GNATS.
 > 
 > From: Martin Husemann <martin%duskware.de@localhost>
 > To: gnats-bugs%netbsd.org@localhost
 > Cc:
 > Subject: Re: port-sparc/58352: ld: /usr/lib/libubsan.so: undefined reference
 >   to `__sync_val_compare_and_swap_8'
 > Date: Wed, 19 Jun 2024 16:51:37 +0200
 > 
 >   On Wed, Jun 19, 2024 at 02:45:02PM +0000, Rin Okuyama wrote:
 >   >  (3) Earliest CPU model satisfy (2) seems `ultrasparc3`, IIUC:
 >   >  (I'm not familiar to sparc although...)
 >   
 >   All ultrasparc CPUs support VIS and can do 64bit atomic ops.
 >   
 >   The ultrasparc3 introduced "VIS 2" with additional instructions.
 
 Hmm, I tried `-mcpu=ultrasparc`, but generated library has reference to
 __sync_val_compare_and_swap_8. I've not figured out why, at the moment.
 
 >   However, for 32bit sparc CPUs we can not assume support (and it is pretty
 >   lame of ubsan to use it). Can we patch the ubsan code to only require 64bit
 >   atomics #ifdef _LP64 ?
 
 Yeah, this is just what I tried. It works at least on i386, but
 diff is *really* large:
 
 https://gist.github.com/rokuyama/24d0ed7e537fffc07e2975c3880817f1
 
 I wonder whether we can ship atomic64 runtime routines with
 lib*san as, e.g., hidden symbols...
 
 Thanks,
 rin